A Realistic Experiment of a Wireless Sensor Network on Board

A Realistic Experiment of a Wireless Sensor Network on
Board a Vessel
Hussein Kdouh, Christian Brousseau, Gheorghe Zaharia, Guy Grunfelder,
Ghaı̈s El Zein
To cite this version:
Hussein Kdouh, Christian Brousseau, Gheorghe Zaharia, Guy Grunfelder, Ghaı̈s El Zein. A
Realistic Experiment of a Wireless Sensor Network on Board a Vessel. 9th International Conference on Communications, COMM 2012, Jun 2012, Bucarest, Romania. pp.189 - 192, 2012.
<hal-00776868>
HAL Id: hal-00776868
https://hal.archives-ouvertes.fr/hal-00776868
Submitted on 16 Jan 2013
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
A Realistic Experiment of a Wireless Sensor Network
on Board a Vessel
H. Kdouh1,*, C. Brousseau2 , G. Zaharia1, G. Grunfeleder1 and G. El Zein1
IETR - Institut d’Electronique et de Télécommunications de Rennes
1
INSA de Rennes, 2Université de Rennes 1
UMR CNRS 6164, Rennes, France
*
E-mail : [email protected]
Abstract—Wireless Sensor Networks (WSN) may be a very
useful technology for monitoring systems in hostile environments.
Few works have treated the use of this technology in the
particular metallic shipboard environment. This paper reports on
the deployment of a WSN on board a ferry-type boat during
realistic conditions. The network was tested during sailings and
stopovers for several days. The results of our previous papers
reporting on the radio wave propagation on board ships are
recalled. The network performance and a comparison of its
evolution with respect to previous results are presented. In spite
of the metallic structure of ferries and the dynamic movement of
crew and passengers on board, the results have shown a very
good network reliability and connectivity. The previous
conclusions have been also confirmed by the topology evolution of
the network and the analysis of RSSI levels of links between
sensor nodes.
Keywords-Wireless sensor network; ships; watertight doors;
I.
INTRODUCTION
Standardized monitoring systems have been designed to
ensure safe public use of structures such as bridges, aircrafts,
ships and pipelines. Ships are one of the most used structures
for commercial and military purposes. Existing shipboard
monitoring system uses extensive lengths of cables to connect
sensors and actuators to central control units. The reduction of
the huge amount of cables using a wireless communication
between sensors and central units would greatly reduce the cost,
weight and architecture complexity of vessels [1]. However,
wireless communication on board ships is limited by several
factors such as metallic bulkheads, watertight doors and
multipath effects. A feasibility study of wireless
communication on board vessels must be conducted to verify
the possibility of replacing the current monitoring system by a
Wireless Sensor Network (WSN). Few papers have studied the
wireless communication on board ships. In [2], the authors have
conducted ZigBee measurements on the passenger deck of a
ship and a small WSN has been deployed between the main
engine room and the control room. In [3], a WSN has also been
tested successfully in the main engine room of a ship. In [4], we
have tested a WSN on board a ferry moored to the harbor. All
these measurements have been carried out when the ships were
moored to the port. To date, experiments during ship operation
have not been carried out. Since the main engine or other
equipment and the passengers’ movements can affect the
quality of wireless communication, it is necessary to conduct
measurements with a ship in operation. This paper investigates
the feasibility of WSN on board vessels during realistic
conditions. Based on the results of our previous studies, we
have deployed a WSN in the three lower decks of a ferry called
“Armorique”. The network has been tested for several days
during sailings and stopovers between Roscoff (France) and
Plymouth (United Kingdom). All realistic conditions (crew's
and passengers' movements), fixed and mobile vehicles
(“mobile” only during stopovers) in the parking, turned on
motors and generators in engine rooms, etc…) have been
considered. Sensing data (temperature, humidity, pressure,
ambient light and acceleration), as well as data packets (sent
and dropped packets, received signal strength indicator (RSSI),
battery voltage) have been gathered and sent by sensor nodes
to a base station placed in the control room.
The remainder of this paper is organized as follows:
Section II describes the considered environments. Section III
recalls the results obtained from our previous measurements on
board “Armorique” vessel. The network setup, deployment
procedure and results of the WSN test are shown in Section
IV. Finally, conclusions are drawn in Section V.
MEASUREMENTS SITES
The test has been carried out on board “Armorique” vessel,
which is a modern ferry from the Brittany Ferries company.
This 29500-tons vessel is characterized by a length of 168.3 m,
a beam of 26.8 m and a capacity of 1500 passengers [5]. The
deckhouse of “Armorique” consists of the followings: the
bottom deck which contains all engine rooms, the second deck
which houses the control room, the third, the fourth and the
fifth decks which are vehicles parking, the sixth and the seventh
decks which contain restaurants, shopping and entertainment
areas, the eighth and the ninth decks which house the
passengers' cabins, and the bridge deck which contains the
wheel house and the crew's cabins.
II.
The tested network has mainly covered a part of the three
lower decks including the parking (deck 3), the control room,
the tank room, the rudder gear room, the changing room and the
workshops room (deck 2), and the main and auxiliary engine
rooms AER (decks 1 and 2). These configurations have been
chosen for several reasons. Firstly, the considered environments
are the most hostile for the wireless propagation on board the
vessel (totally metallic bulkheads and watertight doors).
Secondly, most of the sensors (and the most critical ones) of the
monitoring system are located in these areas. Finally, a WSN
deployed in these areas can simulate all communication
scenarios on board the vessel (communication between
compartments or between decks, door opened or closed...). The
engine rooms of “Armorique” ferry are constituted of
equipment such as engines, generators, valves and pumps. All
equipment and walls in this environment are made of metals,
mainly steel. The parking (deck 3) is a big hall with metallic
walls. A stairway, with two metallic watertight doors on its two
sides, connects the control room and the parking. During 105
hours test, the ferry has made 10 cruises between Roscoff and
Plymouth. After each sailing, the ferry makes a stopover for
discharge, charge and maintenance. During this period, vehicles
enter to or exit from the parking. Crew members were in
continuous movement between rooms and decks containing
sensor nodes, which leads to a frequent doors opening and
closing. This frequent motion on board the ferry changes the
characteristics of the propagation channel and modifies
frequently the quality of the link between nodes.
III.
PREVIOUS RESULTS
Point-to-point measurements have been previously
conducted on board “Armorique” in the ISM 2.4 GHz band
[6, 7]. Different configurations and measurement scenarios
have been analyzed. A communication was considered as
possible if the power of the received signal was higher than the
sensitivity of an IEEE 802.15.4 compliant receiver (-90 dBm)
that will be used in the network test (the transmission power
was 0 dBm). The most important drawn conclusions are the
followings:
1) The link between two nodes placed in the same room is
often possible with a good quality, whatever the
configuration of communication between the
transmitter and the receiver (Line-of-Sight or No Lineof-Sight).
2) The communication is not possible between two nodes
located inside two adjacent rooms which do not have a
common door. Some exceptions may be found when a
huge amount of cables is installed between adjacent
rooms.
3) The communication is possible between two nodes
placed in adjacent rooms that have a common door
(even if the door was closed).
4) The excess path loss due to the closure of a watertight
door is about 25 dB (~17 dB) if the door is near to (far
from) the direct path between the transmitter and the
receiver.
5) The communication is not possible between two nodes
separated by two closed watertight doors.
6) The doors and the cables installed between adjacent
rooms are probably the only sources of radio leakage
for signal propagation between adjacent rooms.
7) The communication between two nodes located in two
adjacent decks is possible only when they are placed
near to a stairway relating the two decks. Stairways are
probably the only radio leakage between adjacent
decks.
IV.
WIRELESS SENSOR NETWORK TEST
This section describes the network setup including the test
equipment and the deployment procedure, presents and
analyses the collected results.
A. Network Setup
The shipboard WSN test was carried out using Crossbow’s
MICAz wireless sensor nodes (motes). Each node has a
maximum data rate of 250 kbps and is equipped by a sensor
board supporting temperature, humidity, barometric pressure,
ambient light and acceleration sensors. Rather than inventing a
new routing protocol, we have decided to apply Crossbow’s
XMesh routing protocol to evaluate its efficiency in such a
rough environment. XMesh is a link-quality based dynamic
routing protocol that uses periodic Route Update (RU)
messages from each node for link quality estimation. Each node
listens to the radio traffic in the neighborhood and selects the
parent that would be the least costly in terms of transmissions
number to reach the base station [8]. Hence, XMesh may be
suitable for the particular dynamic and hostile shipboard
environments where the link quality between nodes is the most
unstable parameter. XMesh may be configured in one of three
power modes: High Power (HP), Low Power (LP) and
Extended Low Power (ELP). As we are dealing with the
problem of network connectivity, we have used the HP mode
with motes always powered. This mode allows fixing a
minimum value of 0.3 seconds to the period of data
transmission by sensor nodes. A huge number of packets are
sent by each sensor node during the network test, which
simulates an emergency case where a large number of sensors
send data frequently and simultaneously to the base station.
Moreover, this mode gives a reliable statistical estimation on
the links quality and the behavior of nodes with respect to the
hostile and dynamic environment.
The deployed WSN is constituted of 25 sensor nodes placed
in decks 1, 2 and 3. As stated before, the placement of nodes is
based on conclusions drawn from point-to-point measurement
campaigns. As hatches are the main radio leakage between
rooms (conclusion 6 in section III), we have placed nodes near
the doors in each room. Moreover, stairways are the main radio
leakage between adjacent decks. Hence, we have placed nodes
in the stairway between the control room and the parking.
Several nodes have been placed at locations containing real
sensors in the tank room, the auxiliary and main engine room
and the parking. These nodes will be called peripheral nodes in
the reminder of this paper. The gateway node was fixed in the
control room in deck 2 and was connected to a laptop via a
USB connection. The laptop was running MoteView which is a
graphical user interface developed by Crossbow Technology to
visualize the real-time sensing data and the network topology
evolution.
B. Sensing Data
As mentioned before, Micaz motes are equipped with
sensor boards supporting temperature, humidity, barometric
pressure, ambient light and acceleration sensors. We present
here a data sample obtained from two sensor nodes placed in
two different rooms. Fig. 1 shows the evolution of vertical
acceleration sensed by node N5 in the tank room and node N19
in the main engine room during the test.
Figure 1. Vertical acceleration in the main engine room (MER) and the tank
room
Blue and red curves represent respectively the data
measured by node N19 and node N5. Vertical acceleration in
the tank room is almost constant during the test. This value
presents some small oscillations during sailings corresponding
to vertical motion of the ship in the sea. However, the
difference between sailings and stopovers is clearer in the main
engine room where other sources of oscillations exist. The
oscillations of engines (which are turned on only during
sailings) lead to the big difference between values of vertical
acceleration in the main engine room.
C. Routing Performance
The results of packets statistics have shown that more than
97% of data packets have arrived successfully to the base
station. These results reflect a very good reliability of the
network and significant routing performance. The results also
indicate that the percentages of forwarded packets were high for
hatches and stairways nodes. This result was expected as these
nodes were placed near to watertight doors and in the stairways,
in order to route sensing data coming from rooms. Among these
routers, the nodes located in the radio range of the gateway
node (critical nodes) have the maximum values of forwarded
packets. This is explained by the fact that all sensor nodes must
select one of these nodes to forward its data to the base station.
In spite of the significant reliability of XMesh with respect
to the transmission ratio, the power consumption does not show
the same performance. We expected that power consumption of
critical nodes will be clearly greater than that of peripheral
nodes. However, the measurements indicate a small difference
between power consumptions. These nodes have similar
number of originated packets, but critical nodes have forwarded
a number of packets 65 times greater than that forwarded by
peripheral nodes. Since the transmission and reception are the
two main factors that increase the power consumption of a
sensor node, critical nodes must have a greater consumption
and they will stop working before the peripheral nodes.
Nevertheless, the results indicate slightly larger power
consumption and shorter lifetime for critical nodes. This
similarity may be explained by the overhearing problem, by
which nodes waste energy in receiving all packets that are
transmitted in their neighborhood, even if they are intended for
others. Another factor may be the small route updates interval
time where sensor nodes consume energy in collecting health
data about their neighbors. A power control mechanism must be
developed in order to enlarge the shipboard WSN lifetime. As
sensor nodes send their sensing data mainly via hatches nodes
or stairway nodes (critical nodes) and the links established have
shown good stability and quality, it will be useful to reduce the
route updates frequency to save energy. Moreover, critical
nodes supporting a very high traffic may be powered by the
mains supply.
D. Selection of Parent Nodes
In a multi-hop routing protocol, a parent node is defined as
the node selected by a sensor for the next hop in order to send
its data packets to the base station. As previously stated, sensor
nodes are preprogrammed by the XMesh routing protocol.
Therefore, a sensor node selects the next hop which minimizes
the transmissions needed to send a packet to the base station.
Hence, the selection of parent nodes is based on the link quality
between a sensor node and its neighbor nodes, and the duration
of this parent-child connection is based on the link quality. We
have compared the parent list of sensor nodes with the
conclusions of Section III. Preliminary analysis of the results
shows a good agreement between the established parent-child
links and the previous conclusions. Hatches and stairway nodes
have been often selected as parent nodes for peripheral nodes.
However, “strange links” between some nodes separated by
two watertight doors have been detected. The fifth conclusion
says that the communication becomes impossible when two
watertight doors between the two communicating nodes are
closed. Therefore, we have studied the evolution of RSSI of
these “strange links” to determine if the two watertight doors
were really closed when the links have been established. Fig. 2
shows an example of this case. Node N3 has selected the
gateway node N0 as its parent node for 76 % of packets and
nodes N1 and N12 for the remaining packets. The
establishment of links N3-N1 and N3-N12 is coherent since the
two nodes of each link are separated by only one watertight
door (which is consistent with the third conclusion given in
Section III). However, following the fifth conclusion in Section
III, the connection between N3 and N0 is not possible if both
the watertight door of the tank room and the door of the control
room are closed.
Fig. 3 shows the evolution of RSSI levels between N3 and its
parent nodes during the test. It can be seen that N3 has selected
N0 as parent node only between instants t = 12 and t = 99 and
N1 or N12 for the remaining time. Most of RSSI values of link
N3-N1 are grouped around a mean value of -80 dBm. Some
values increase to -55 dBm after t = 12. N1 and N3 are
separated by one door located in the direct path between them.
The gap of 25 dB between the two RSSI levels of link N3-N1
represents the excess path loss due to the closure of this door.
Therefore, the RSSI level of N3-N1 can be used as an indicator
of the door state (open/closed). Hence, we can simply notice
that N3 has switched to N0 when the tank room door was open.
deviations of RSSI levels of parent-child links were high, which
is simply explained by the dynamic behavior of the
environment. However, LS is less than 7% for most of nodes,
which reflects significant link stability. The reflective walls
leading to multiple paths between nodes may explain the
stability of links even dynamic link quality fluctuations.
V.
Figure 2. N3 and its parent nodes
The link N3-N0 remains established until t = 100 when the door
becomes closed. Therefore, when N0 was the parent node of
N3, only one door (control room door) was closed and N3 has
switched to N1 or N12 when the second door has been closed.
Thus, this result is consistent with the conclusion 5 of Section
III. It is also worth mentioning that most of RSSI values of link
N3-N0 are grouped around a mean value of -83 dBm and some
values increase to -65 dBm. This gap of 18 dB is due to the
opening of the control room door which is placed far from the
direct path between N3 and N0 (this result is also consistent
with the fourth conclusion of Section III). We have followed
the same procedure to verify that all links between two nodes
separated by two doors were established if at least one door was
open.
CONCLUSION
In this paper, we have investigated the performance of a
WSN on board a ferry-type ship. The objectives were to study
the effect of realistic circumstances on the network
performance and to compare the network behavior with our
previous results obtained from propagation studies. The
deployed network has been tested under realistic circumstances
during five days. The results have shown a significant network
reliability and robustness in spite of the dynamic and hostile
covered environments. Moreover, the parent nodes selection
was consistent with the previous conclusions and the same
watertight door attenuation has been found from the values of
RSSI. This test has also shown that link-based routing protocols
may be a reliable solution for future shipboard WSN.
However, some enhancements must be carried out on the power
consumption mechanism. The results have indicated good links
quality and stability, which allows reducing the frequency of
routes updates messages communication. Briefly, we proved in
this paper the possibility to reduce the amount of wires of the
shipboard monitoring system by using the WSN technology.
ACKNOWLEDGMENT
The authors would like to thank Brittany Ferries and
Marinelec Technologies companies for the opportunity of
conducting the measurements on board the “Armorique”, and
the “Pôle Mer Bretagne” and the Council of Brittany (“Région
Bretagne”) for their financial support.
REFERENCES
[1]
[2]
Figure 3. RSSI vs Time of links between N3 and its parent nodes
E. Link Quality and Link Stability
The analysis of data packets sent by each sensor node also
concerns the “retries”, determining the number of times a node
had to retransmit a packet due to lack of link-level
acknowledgment. This number helps us to estimate the link
quality between a sensor node and its parent node, as a
percentage of successful transmissions with respect to the
number of all transmissions. We have studied the link quality of
each sensor node with all its parent nodes. All motes have a
mean value of link quality greater than 70%. Critical nodes
have a mean value higher than 90%, in spite of the huge traffic
that they have supported and the dynamic environment around
the control room. Moreover, we have estimated the link
stability (LS) of a node as the percentage of parent changes
with respect to the total number of node packets. The standard
[3]
[4]
[5]
[6]
[7]
[8]
J. P. Lynch and K. J. Loh, “A summary review of wireless sensors and
sensor networks for structural health monitoring”, The Shock and
Vibration Digest, 38 (2), pp.91-128, 2006.
P. Bu-Geun, C. Seong-Rak, P. Beom-Jin, L. Dongkon, Y. Jong-Hwui,
and B. Byung-Dueg, “Employment of wireless sensor networks for
fullscale ship application”, In Proc. Int. Conf. Embedded and Ubiquitous
Computing (EUC 2007), pp.113-122, 2007.
T. Pilsak, T. Schröder, J. Eichmann and J. L. ter Haseborg, “Field test of
a wireless sensor network inside the engine room of a vessel”, Hamburg
University of Technology Institute of Measurement Technology, 2009.
H. Kdouh, C. Brousseau, G. Zaharia, G. Grunfelder and G. El Zein,
“Applying of ubiquitous wireless technologies for shipboard monitoring
systems”, 14th international symposium on wireless personal multimedia
communication, Brest, France, Oct. 2011
http://www.brittany-feeries.co.uk
H. Kdouh, G. Zaharia, C. Brousseau, G. Grunfelder, G. El Zein,
“ZigBee-based sensor network for shipboard environments”, ISSCS
Romania, July 2011.
H. Kdouh, C. Brousseau, G. Zaharia, G. Grunfelder and G. El Zein,
“Measurements and path loss models for shipboard environments at
2.4 GHz”, European Microwave Conference EuMC 2011, Manchester,
UK, 2011.
Memsic
Technology,
“Xmesh
user’s
manual”,
http://www.memsic.com/support/documentation/wirelesssensor-networks/category/6-user-manuals.html